An engineering classroom that thinks before it consumes.
A student-built exploration of spatial sensing, environmental intelligence, measurable power control and human-first automation — designed for demonstration, testing and real-time operation.
The interface treats the project as shared team knowledge: every member should be able to explain the problem, sensing, control logic, power allocation, limitations, safety and measured evidence — not only an individual sub-task.
01
SN
Shanmugam N
Roll No. 106126118
02
GAJ
Gargi Anant Jirapure
Roll No. 106126042
03
APS
Aditya Pratap Singh
Roll No. 106126008
Website & interactive system experience04
RM
Ritesh Mannem
Roll No. 106126114
05
YAG
Yashika Anvi Gopu
Roll No. 106126142
06
RKS
Ram Karthik S
Roll No. 106126110
07
AKK
Abhay Karthik Krishna
Roll No. 106126002
08
MM
Mudireddy Mukesh
Roll No. 106126078
09
KKM
Ksshitiz Kumar Mittal
Roll No. 106126064
TEAM PRINCIPLE
One source of truth. One control philosophy. One consistent explanation from every member.
F1.0 ENGINEERING BLUEPRINTLOCAL-FIRST CONTROL
यत्र आवश्यकता, तत्र ऊर्जा।
Energy that understands the classroom.
A three-zone adaptive classroom that senses occupancy, daylight and environmental demand—then decides which loads deserve power inside a measured operating budget. Each decision remains visible, so the audience can follow what the system sensed, why a priority changed, what was allocated and what electrical effect was measured.
03independent zones
05decision stages
100%local core control
0fabricated savings claims
VOLUMETRIC CORELIVE DECISION ORBITS
SENSEPIR · LDR · SCD30
DECIDEESP32 · Energy Manager
VERIFYINA219 · Measured Power
ऊर्जाA continuous decision loop. The rotating volumetric nucleus represents the local ESP32 control core, while electrons travel on genuinely projected 3D orbital planes around it—visually separating front, back and depth as live inputs, changing priorities and measured feedback circulate through the system. The motion is deliberately continuous because the classroom does not make one decision once—it repeatedly senses, evaluates, allocates and verifies.
Connect the browser to the ESP32 over USB Web Serial or a local WebSocket. Until a hardware link is active, every value is explicitly marked as simulation.
SIMULATION MODEDemo telemetry only — not a live measurement.
3D CLASSROOM / LIVE STATE
SIMULATED
3D ENGINE LOADINGDrag to orbit · wheel to zoom · click a zone in simulation
PIR fieldlightingairflowenvironment
01FRONT
02MIDDLE
03REAR
VENT
Simulation tip: click a zone to toggle occupancy. In LIVE mode the room becomes read-only and mirrors the ESP32.
SENSOR / SYSTEM DEGRADEDNo active faults.
HARDWARE BRIDGE
USB live link + optional LAN bridge.
The included final firmware connects through USB Web Serial in Chromium browsers. The same telemetry schema can also be used by a future/custom WebSocket bridge, but that LAN server is not bundled in the supplied firmware.
Advanced/optional: the included final ESP32 firmware uses USB Web Serial; it does not create this WebSocket server by itself. Use this field only if you later add a compatible LAN bridge. HTTPS pages also block insecure ws://; use localhost/http or a secure wss:// bridge.
No hardware connectedSimulation engine is active.
TELEMETRYLast packet: —COMMANDNo pending command
SYSTEM MODE
Physical override remains the authority in the real prototype. Dashboard manual controls are isolated in a clearly labelled service mode.
ENERGY BUDGET DIAL100%
40% constrained100% full budget
Reduce the permitted controllable-load budget and watch lower-priority flexible loads yield first.
01.5 / INTERACTIVE SYSTEM LAB
Touch every input. Watch every decision.
This lab is deliberately separated from measured evidence. In SIMULATION MODE you can change occupancy, daylight, CO₂, temperature, humidity and load budget to understand the control logic. When real hardware is connected, sensor controls become read-only and the ESP32 remains the authority.
SIMULATION ≠ MEASUREMENTUse this section to explain logic and rehearse the demo. Do not present these generated values as experimental results.
SPATIAL INPUTSOccupancy + daylight
Zone 1 · Front
Zone 2 · Middle
Zone 3 · Window side
The LDR values are normalized relative brightness, not precision lux. Each zone is calibrated separately.
ENVIRONMENTAL INPUTRoom-level demand
<1000 ppm · normal1000–1500 · ventilate>1500 · high priority
These control bands are prototype demonstration setpoints. They are not presented as universal health or legal limits.
GUIDED DEMOFaculty-facing sequence
PRESENTER LINE
Choose a step. The system will change state and show the exact engineering point you should explain.
MANUAL SERVICEExplicit, logged intervention
Manual commands are meaningful only in MANUAL SERVICE. In the final firmware, the physical selector and safety rules must still have higher authority than the browser.
02 / THE PROBLEM
A classroom is dynamic. Its electrical control usually isn’t.
Students cluster unevenly. Daylight is different near windows. Thermal and ventilation demand change during the same lecture. A single wall switch sees none of it.
01
All-or-Nothing
One occupied region causes the whole room to operate. Room-level control ignores spatial demand.
◫02
Empty-Zone
Lights or fans can operate where nobody is sitting—power is consumed without providing service.
○03
Daylight Blindness
Artificial light stays high even when daylight is sufficient. Natural light is not treated as an energy resource.
☼04
Comfort Blindness
Presence alone cannot tell whether occupied air is warm, humid or in need of greater ventilation.
≈05
Static Allocation
Loads switch independently. No mechanism decides which demand deserves priority when capacity is constrained.
⇄06
Unverified Savings
Switching something off is not proof of energy saved. The project measures voltage, current and real power.
∿
THE DESIGN SHIFT
“From automatic switching to energy decision-making.”
Human need before automation · local-first control · measured claims · low-voltage safety · explainable decisions · graceful failure.
03 / SYSTEM ARCHITECTURE
Every signal earns its action.
Inputs do not directly switch outputs. Raw sensing becomes interpreted demand; demand is prioritized; the energy manager allocates; measured power closes the evidence loop.
→→→→
SPATIAL INPUTPIR ×3Zone occupancy
SPATIAL INPUTLDR ×3Zone daylight
ENVIRONMENTSCD30CO₂ · Temp · RH
FEEDBACKINA219Voltage · Current · Power
LOCAL CONTROL BRAINESP32Sensor Manager Decision Engine Energy ManagerWi-Fi + I²C + ADC + PWM
ACTUATIONLIGHTS ×3PWM channels
ACTUATIONZONE FANS ×3Local circulation
PRIORITY LOADVENTILATIONAir-quality response
INTERFACEOLED + WEBState · reason · evidence
04 / SPATIAL INTELLIGENCE
Three zones. Three local realities.
Three zones are complex enough to demonstrate front/middle/rear variation without turning a student prototype into a wiring maze.
1
Zone 1 · Front
Motion history decides logical occupancy. Local brightness decides whether artificial light is actually needed.
PIR occupancy input
LDR normalized brightness
Independent LED channel
Local fan channel
2
Zone 2 · Middle
An empty middle zone can remain unpowered even while students occupy the front or rear of the classroom.
Software occupancy hold
Hysteresis prevents chatter
Filtered ADC values
Priority-aware allocation
3
Zone 3 · Window side
Local daylight can reduce only the lighting that became unnecessary instead of dimming the entire room.
Window-side daylight demo
Independent calibration
Reduced / minimum light
Other zones unaffected
IFEMPTY→Light OFF · Fan OFF
IFOCCUPIED + DARK→High light request
IFOCCUPIED + BRIGHT→Reduced / minimum light
05 / ENVIRONMENTAL INTELLIGENCE
Presence says where. Environment says what is needed.
A representative room-level SCD30 adds CO₂, temperature and relative humidity. These are converted into transparent demand states—not a mysterious “AI score”.
CO₂
Ventilation priority
< 1000 ppm normal · 1000–1500 ppm ventilate · >1500 ppm high priority. These are configurable demonstration setpoints, not universal health limits.
°C
Thermal demand
<27 °C normal · 27–30 °C warm · >30 °C hot. Temperature increases fan priority only for occupied zones.
RH
Context, not false control
Relative humidity is monitored and flagged. The project does not claim that a simple fan “controls” humidity.
SCD30
MEASUREMENT QUALITY
Placement is part of the sensor.
Keep the environmental sensor away from direct fan discharge, lamp heat, the ESP32 regulator and the window opening. A breath challenge can demonstrate response—but must never be presented as a normal room-average measurement.
06 / DYNAMIC ENERGY REALLOCATION
Not energy transfer. Load-budget intelligence.
When lower-priority demand falls, a higher-priority load may use more of the same permitted budget. Nothing is physically “moved” from a lamp to a fan.
INTERACTIVE ALLOCATION LAB
Turn the budget down.
Reserve minimum service first, then allocate the remaining budget by priority. The visual below follows the same logic as the live command centre.
P1High CO₂ → ventilation
P2Occupied + hot → zone fan
P3Occupied + dark → minimum light
P4Normal occupied service
P5Optional / low-value load
PERMITTED BUDGET100%
For a visible priority demonstration, prepare the high-demand case first. It creates three occupied zones, mixed daylight, high CO₂ and a hot room in SIMULATION MODE.
Requested —Protected minimum —Allocated —State NORMAL
DON’T SAY“We transfer saved electricity from a light to a fan.”SAY“Reduced low-priority demand makes more of the permitted shared load budget available for a higher-priority load.”
07 / BUILD JOURNEY · 0 → WORKING MODEL
Build it in layers. Never debug everything at once.
This section converts the blueprint into a practical assembly sequence. Select a stage to see exactly what is being built, why that stage exists, what to verify before moving forward and what failure would look like.
01
POWER FOUNDATION
Bench-test the power architecture
WHAT YOU DO
WHY IT EXISTS
PASS BEFORE NEXT STAGE
COMMON FAILURE
1 / 12
08 / PHYSICAL COMPONENT EXPLORER
Every part has a reason to exist.
Choose any physical item to understand its quantity, role, connection, reason for selection, what the software expects from it, how it should be mounted and the mistake most likely to damage the demo.
BASELINE ARCHITECTURE3 zones · 7 controllable load channels · 1 local controller · 1 room environmental node
SAFETY BOUNDARYLow-voltage DC prototype only. No direct 230 V student wiring.
09 / HARDWARE CONSTELLATION
Every component has one explainable job.
The prototype remains low-voltage DC. ESP32 pins are control signals—not power outputs—and no student breadboard wiring is connected directly to 230 V AC.
CONTROL BRAINESP32ADC · I²C · PWM · Wi‑Fi
09.5 / INTERACTIVE WIRING LAB
From pin number to physical current path.
Click any signal or power block. The diagram separates sensor inputs, I²C, controller outputs and the protected 12 V / 5 V / 3.3 V rails so a team member can explain exactly what connects where and why.
ZONE INPUTS · ADC1 / DIGITAL
LOCAL BRAINESP32DevKitC / WROOM class
I²C BUSSDA 21 · SCL 22
PHYSICAL OVERRIDEGPIO 13 / 14
CONTROL OUTPUTS · MOSFET DRIVERS
12 V ADAPTER→ 3 A fuse → INA219 → controllable load rail
Each subsystem updates on its own schedule. A slow environmental sensor must never stop an override read, power check or zone update.
SensorManagerPIR · ADC · SCD30 · INA219 · modescheduled by sensorFilterManagerdebounce + smooth ADCeach sampleOccupancyManagerlast-motion state100–250 msLightingManagerbrightness + occupancy → demand250–500 msEnvironmentManagerCO₂ / temp / RH states~2 sEnergyManagerpriority + minimums + budget250–500 msActuatorManagerPWM + fan commandson changePowerFeedbackmeasured V / I / P correction250–1000 msUIManagerOLED + web state500–1000 msLoggersamples + state transitions1–5 s + events
MAIN LOOP / EXPLAINABLE PSEUDOCODE
11 / EVIDENCE
Measure first. Claim later.
The dashboard’s third job—after monitoring and explanation—is evidence. Logged sensor state, allocation and measured watts must agree with every reported result.
POWER TRACE
RequestedAllocatedMeasured / budget basis
EVENT LOG
TIMEtimestamp / elapsed
OCCUPANCYZ1 · Z2 · Z3
LIGHT INPUTraw + normalized
ENVIRONMENTCO₂ · temp · RH
DEMANDrequested loads
ALLOCATIONfinal commands
POWERV · I · W · Wh
EVENTSfaults · transitions
12 / VALIDATION LAB
Ten test cases. No hand-waving.
In simulation mode these buttons demonstrate expected logic. Formal project evidence still comes only from the fabricated model and recorded measurements.
13 / FAIL-SAFE ENGINEERING
Detect. Protect. Fall back. Record. Recover.
A failed sensor should degrade one feature rather than collapse the entire system. Override and fault states are first-class design states—not emergency code appended later.
DETECT problem→PROTECT user/system→FALL BACK safely→RECORD what happened→RECOVER re-test
PIR limitation
A completely still occupant may eventually be missed. Hold time and override manage the prototype; presence-grade sensing is the real upgrade.
One SCD30
One representative room-level sensor cannot describe every gradient in a large real classroom.
Relative daylight
LDRs demonstrate normalized brightness, not certified lux measurement.
Model ≠ building
Mini fans and LED strips prove control behaviour, not full-scale airflow, photometry or mains compliance.
LOW-VOLTAGE BOUNDARYDo not connect the ESP32, breadboard or student wiring directly to 230 V AC. Real deployment requires rated dimming/relay hardware, isolation, protection and qualified installation.
14 / SCALE RESPONSIBLY
Local autonomy. Campus-level visibility.
Each classroom can remain autonomous while publishing summary data upward. Network failure may remove central visibility, but it should not remove local classroom function.
CAMPUSFederated energy intelligenceAutonomous rooms · central analytics
15 / PROJECT ATLAS · 0 → END
The complete project, one chapter at a time.
This is the long-form study layer. Choose any chapter on the left, then use the tabs to open the full explanation, engineering reasoning, faculty-facing wording and viva / caution notes. It follows the F1.0 blueprint rather than inventing a second version of the project.
00
Blueprint pages 1–5
START HERE
Project identity & evidence rules
REMEMBER
1 / 19
16 / PRESENTATION MODE
Explain the whole project without losing the room.
INTERACTIVE EXPERIENCE CREDIT
Designed to make engineering visible.
Developed by Aditya Pratap Singh
Roll No. 106126008✦CSE✦1st Year Undergraduate✦NIT Tiruchirappalli
This website is the interactive presentation, simulation and live-dashboard layer for the Smart Classroom project. The physical ESP32 remains the control authority; the interface translates its sensing, decisions, priorities, power allocation and measured feedback into a form that can be followed step by step. The goal is not only to make the prototype look intelligent, but to make every important decision understandable to the person standing in front of it.